Artificial intelligence intervenes in labor not as a sector-specific technology, but as a force of systemic reordering of the global economy. Its impact is not measured solely in terms of task automation or productivity growth, but in the reconfiguration of the relationships between space, value, and decision. In this sense, AI acts as a geographic technology of economic power, capable of shifting the center of gravity of labor from visible and localized activities toward immaterial, concentrated, and hardly transferable functions.
The new hierarchies that emerge no longer follow the traditional line of fracture between industrialized countries and low-cost labor areas. According to estimates by the International Labour Organization, more than 60 percent of the value added generated by advanced artificial intelligence systems is today concentrated in a limited number of economies that combine computational infrastructures, massive access to data, and the capacity for strategic integration of these resources. Economic power thus tends to shift toward those who control the decision-making nodes of global chains, rather than toward those who host the largest volumes of employment.
Productive value migrates along increasingly intangible trajectories. While globalization over recent decades had delocalized labor by pursuing wage differentials, the algorithmic economy reorients the value chain upstream, toward the phases of model design, definition of decision parameters, and supervision of automated systems. Executory activities, which had constituted the employment base of many emerging economies, are progressively absorbed by automated processes or standardized to the point of losing bargaining power. According to the McKinsey Global Institute, by 2030 up to 30 percent of hours worked globally could be automated, yet more than 70 percent of the value generated will remain concentrated in functions with high cognitive and decision-making intensity.
This recomposition does not simply produce less work, but structurally different and more polarized work. On one side, a limited share of high-value occupations linked to the design, governance, and interpretation of intelligent systems; on the other, a growing mass of residual, intermittent, or platform-dependent activities, lacking control over the criteria that determine their remuneration. This is a dynamic that many multinational firms already experience internally, with the concentration of strategic functions in a few global hubs and the peripheralization of entire operational areas.
At the macroeconomic level, this transformation introduces a new form of inequality among states. Economies that fail to develop autonomous capabilities in the field of AI risk sliding into a condition of cognitive dependence, in which local labor is integrated into decision-making systems designed elsewhere. This is no longer merely technological dependence, but normative subordination: those who do not control the models do not participate in defining success metrics, efficiency criteria, and allocative priorities. The World Bank has recently highlighted how countries with weaker advanced digital capacity tend to lose bargaining power even within traditional value chains, precisely because they are unable to influence automated decision-making phases.
The social consequences of this reordering go beyond the employment dimension. Labor has historically functioned as the primary mechanism of social integration and recognition. When AI redefines who works, where, and with what economic weight, it places this implicit pact under strain. Fractures emerge not only between employed and unemployed, but between those included in decision-making circuits and those excluded from them, despite continuing to provide essential inputs. It is a fracture that tends to remain invisible, masked by system efficiency and the apparent continuity of productive flows.
The reflection of Hannah Arendt offers a useful interpretative key to grasp the scope of this change. In her distinction between labor, work, and action, Arendt places labor within the sphere of necessity, while action represents the space of freedom and political responsibility. Advanced automation promises to reduce the burden of necessary labor, but without a redefinition of economic and social institutions, this reduction does not automatically translate into greater space for action. On the contrary, it can produce exclusion, marginality, and loss of voice for large segments of the active population.
In the corporate world, this tension translates into a problem of systemic leadership. Governing globally distributed organizations, in which artificial intelligence mediates a growing share of operational decisions, requires an awareness that goes beyond performance optimization. Companies that have adopted large-scale predictive models have experienced how local efficiency can generate global instability, when algorithmic choices compress decision diversity and amplify internal asymmetries. Leadership, in this context, does not consist in delegating to systems, but in overseeing their distributive and reputational impact.
Artificial intelligence thus acts as a deep current that redraws the geography of labor without appearing directly. States, firms, and communities move within these currents, often reacting rather than governing. Labor ceases to be a stable locus and becomes a mobile configuration, continuously renegotiated according to computational infrastructures and opaque decision criteria. In this scenario, economic value, institutional power, and social cohesion tend to overlap, making it increasingly difficult to separate the question of labor from that of sovereignty and collective responsibility.
Global AI Observatory